ARTICLE

Low concentration IL-1β promotes islet amyloid formationby increasing hIAPP release from humanised mouse islets in vitro

Andrew T. Templin1& Mahnaz Mellati1 & Daniel T. Meier1 & Nathalie Esser1 & Meghan F. Hogan1

&

Joseph J. Castillo1& Rehana Akter2 & Daniel P. Raleigh2

& Sakeneh Zraika1 & Rebecca L. Hull1 &

Steven E. Kahn1

Received: 1 April 2020 /Accepted: 5 June 2020# This is a U.S. government work and not under copyright protection in the U.S.; foreign copyright protection may apply 2020

AbstractAims/hypothesis Aggregation of the beta cell secretory product human islet amyloid polypeptide (hIAPP) results in islet amyloiddeposition, a pathological feature of type 2 diabetes. Amyloid formation is associated with increased levels of islet IL-1β as wellas beta cell dysfunction and death, but the mechanisms that promote amyloid deposition in situ remain unclear. We hypothesisedthat physiologically relevant concentrations of IL-1β stimulate beta cell islet amyloid polypeptide (IAPP) release and promoteamyloid formation.Methods We used a humanised mouse model of endogenous beta cell hIAPP expression to examine whether low (pg/ml)concentrations of IL-1β promote islet amyloid formation in vitro. Amyloid-forming islets were cultured for 48 h in the presenceor absence of IL-1β with or without an IL-1β neutralising antibody. Islet morphology was assessed by immunohistochemistryand islet mRNA expression, hormone content and release were also quantified. Cell-free thioflavin T assays were used to monitorhIAPP aggregation kinetics in the presence and absence of IL-1β.Results Treatment with a low concentration of IL-1β (4 pg/ml) for 48 h increased islet amyloid prevalence (93.52 ± 3.89% vs43.83 ± 9.67% amyloid-containing islets) and amyloid severity (4.45 ± 0.82% vs 2.16 ± 0.50% amyloid area/islet area) in hIAPP-expressing mouse islets in vitro. This effect of IL-1β was reduced when hIAPP-expressing islets were co-treated with an IL-1βneutralising antibody. Cell-free hIAPP aggregation assays showed no effect of IL-1β on hIAPP aggregation in vitro. Lowconcentration IL-1β did not increase markers of the unfolded protein response (Atf4, Ddit3) or alter proIAPP processing enzymegene expression (Pcsk1, Pcsk2, Cpe) in hIAPP-expressing islets. However, release of IAPP and insulin were increased over 48 hin IL-1β-treated vs control islets (IAPP 0.409 ± 0.082 vs 0.165 ± 0.051 pmol/5 islets; insulin 87.5 ± 8.81 vs 48.3 ± 17.3 pmol/5islets), and this effect was blocked by co-treatment with IL-1β neutralising antibody.Conclusions/interpretation Under amyloidogenic conditions, physiologically relevant levels of IL-1β promote islet amyloidformation by increasing beta cell release of IAPP. Neutralisation of this effect of IL-1β may decrease the deleterious effects ofislet amyloid formation on beta cell function and survival.

Keywords hIAPP . IL-1β . Islet amyloid . Type 2 diabetes

AbbreviationsFmoc 9-FluorenylmethyloxycarbonylhIAPP Human islet amyloid polypeptideIAPP Islet amyloid polypeptidemIAPP Mouse islet amyloid polypeptideNLRP3 NACHT, LRR and PYD

domains-containing protein 3RIP Rat insulin promoterTFA Trifluoroacetic acidUPR Unfolded protein response

Electronic supplementary material The online version of this article(https://doi.org/10.1007/s00125-020-05232-2) contains peer-reviewedbut unedited supplementary material, which is available to authorisedusers.

* Steven E. Kahnskahn@uw.edu

1 Division of Metabolism, Endocrinology and Nutrition, VeteranAffairs Puget Sound Health Care System (151) and University ofWashington, 1660 S. Columbian Way, Seattle, WA 98108, USA

2 Department of Chemistry, Stony Brook University, StonyBrook, NY, USA

https://doi.org/10.1007/s00125-020-05232-2

/ Published online: 29 July 2020

Diabetologia (2020) 63:2385–2395

Introduction

Islet amyloid formation occurs in most individuals with type 2diabetes and is associated with beta cell dysfunction and loss[1–5]. Human islet amyloid polypeptide (hIAPP, also knownas amylin) is a beta cell secretory product co-released withinsulin [6] and the amyloidogenic constituent of islet amyloiddeposits [7–9]. Because the sequence of hIAPP and mouseislet amyloid polypeptide (mIAPP) differ in a few criticalamino acid residues, hIAPP is amyloidogenic and cytotoxicwhile mIAPP is not [10]. Consequently, a number of mousemodels with beta cell hIAPP expression have been developedto facilitate the study of islet amyloidosis in type 2 diabetes[11–13]. Most of these models use transgenic expression ofhIAPP under the rat insulin promoter (RIP), but humanisedmice that express hIAPP under the endogenous mIAPPpromoter have also been developed to more accurately modelhIAPP expression and islet amyloid formation [14, 15].

Although the relationship between hIAPP aggregation andbeta cell loss is well-established [1, 16, 17], the mechanismsthat promote cytotoxic hIAPP aggregation in situ remainunclear. hIAPP aggregation has previously been shown toactivate the NACHT, LRR and PYD domains-containingprotein 3 (NLRP3) inflammasome and elicit production ofIL-1β from macrophages and dendritic cells, leading to isletinflammation [18–20]. This production of IL-1β has beenviewed predominantly as a downstream mediator ofamyloid’s cytotoxic effect. More recently, IL-1β signallinghas been proposed as a promotor of amyloid formation. Inhuman islets, pharmacological concentrations of IL-1βincrease amyloid deposition in an IL-1 receptor-dependentmanner [21] and neutralisation of endogenous islet IL-1βwithan antibody has been shown to mitigate islet amyloid deposi-tion in human islets [22]. Similar observations have beenmade in mouse islets that express hIAPP under the RIP [23].However, the mechanisms linking islet IL-1β signalling andamyloid deposition remain to be fully established.

It is difficult to know the concentration of IL-1β the betacell is exposed to in vivo but it is believed that IL-1β concen-trations in the pg/ml range more accurately replicate the isletmicroenvironment [24, 25] than the ng/ml concentrations thatpredominate in the literature [21, 25, 26]. Physiologically rele-vant levels of IL-1β in the pg/ml range have been shown tostimulate insulin secretion [27–30] but the effect on isletamyloid polypeptide (IAPP) release has not been evaluated.Therefore, we hypothesised that pg/ml concentrations of IL-1β directly stimulate release of IAPP from beta cells to furtherpromote cytotoxic hIAPP aggregation under conditions ofislet amyloid formation. In this study, we used a humanisedmouse model with hIAPP expression driven by the endoge-nous mIAPP promoter to examine whether physiologicallyrelevant low pg/ml concentrations of IL-1β increase isletamyloid deposition in vitro and, if so, whether increased

release of hIAPP from the beta cell may be one mechanismunderlying this effect.

Methods

Animals C57BL/6NHsd mice expressing amyloidogenichIAPP under the endogenous mIAPP promoter [14] wereobtained from N. L. Eberhardt (Mayo Clinic, Rochester,MN, USA). Following approval by the Institutional AnimalCare and Use Committee, mice were housed and bred in aspecific-pathogen-free vivarium at VA Puget Sound HealthCare System with ad libitum access to food and water.

Islet isolation and culture Islets from 8- to 12-week-old maleand female mice were isolated by collagenase digestion, thenrecovered overnight in RPMI 1640 medium containing 10%FBS, 1 mmol/l sodium pyruvate, 100 U/ml penicillin,100 μg/ml streptomycin and 11.1 mmol/l glucose. Islets werethen distributed into randomised islet pools, and cultured inmedia containing 16.7 mmol/l glucose and the appropriateexperimental condition for 48 h. The following reagents wereused during the 48 h culture period for each condition: control,BSA (no. A0281; Sigma, St Louis, USA; 4 pg/ml); IgG, goatIgG isotype control (no. AB-108-C; R&D Systems,Minneapolis, MN, USA; 0.4 μg/ml); IL-1β, recombinantmouse IL-1β (no. CYT-273; ProSpec, Rehovot, Israel;concentrations as indicated); NAb, IL-1β neutralising anti-body (no. AF-401-NA; R&D Systems; 0.4 μg/ml).

Histology and quantitative microscopy Islets were fixed in10% (wt/vol.) neutral buffered formalin, embedded in agarand then paraffin, processed and 10 μm sections were cut.Sections were incubated with mouse monoclonal anti-insulinantibody (no. I2018; Sigma; 1:5000) followed by goat anti-mouse Cy3 (no. 115-165-146; Jackson ImmunoResearch,West Grove, PA, USA; 1:250) to visualise beta cells. To visu-alise amyloid deposits, sections were stained with ThioflavinS (0.5% vol./vol. in water) for 2 min at room temperaturefollowed by rapid differentiation in 70% ethanol (10 dips),rapid water rinses (10 dips) and a 5 min wash in water tospecifically stain amyloid deposits and minimise background.Sections were mounted with polyvinyl alcohol containingHoechst 33258 to visualise nuclei [31]. To visualise dead betacells, sections were incubated with anti-insulin antibody (no.I2018; Sigma; 1:5000) followed by goat anti-mouse AlexaFluor 488 (no. A-11001, Life Technologies, Carlsbad, CA,USA; 1:1000), then propidium iodide (no. P3566;Invitrogen, Carlsbad, CA, USA; 1:250) and RNase A (no.R4642; Sigma; 1:300).

Islet area, beta cell area, amyloid area and propidiumiodide-positive nuclei were imaged with a Nikon E800 micro-scope and 20× Plan Fluor objective (Nikon, Tokyo, Japan)

2386 Diabetologia (2020) 63:2385–2395

using an X-Cite 120PCQ fluorescence lamp illuminator(Excelitas Technologies, Waltham, MA, USA). For amyloidimaging, an excitation λ of 488 nm was used (excitation 460–500 nm bandpass filter cube; emission 505 dichromatic mirrorand 510–560 nm bandpass barrier filter). Images were quan-tified using computer-based quantitative imaging software(NIS Elements AR 4.30.02, Nikon, Tokyo, Japan) with theobserver blinded to culture condition of the islets. A mean of26.5 ± 1.1 islets per replicate per condition were analysed.

Preparation and purification of hIAPP hIAPP was synthesisedusing 9-fluorenylmethyloxycarbonyl (Fmoc) chemistry with aCEM Liberty Blue microwave peptide synthesizer (CEMCorp, Matthews, NC, USA) on a 0.10 mmol scale. 5-(4′-Fmoc-aminomethyl-3′,5-dimethoxyphenol) valeric acid(PAL-PEG) resin was used to provide an amidated C termi-nus. hIAPP produced in vivo is amidated at the C terminusand the chemical nature of the C terminus, amide vs carboxyl,has been shown to impact hIAPP aggregation kinetics andinduction of physiological responses [32, 33]. Pseudoprolinederivatives were used at residue 9 and 10 (Ala and Thr), resi-due 19 and 20 (Ser and Ser), and residue 27 and 28 (Leu andSer) to minimise aggregation during synthesis [34, 35]. Thepeptide was released from the resin and the side chainprotecting groups were removed using a trifluoroacetic acid(TFA)-based cleavage cocktail (92.5% TFA, 2.5%triisopropylsilane, 2.5% 3,6-dioxa-1,8-octanedithiol and2.5% water). Crude product was dissolved in 20% acetic acid(4 mg/ml) frozen in liquid nitrogen and lyophilised to increasetheir solubility. The disulfide bond was formed by DMSOoxidation (10 mg/ml peptide in 100% DMSO) at roomtemperature for 4 days. hIAPP was purified using reverse-phase HPLC with a C18 25 mm× 250 mm column (HigginsAnalytical, Mountain View, CA, USA). A binary gradient ofbuffer A (100% H2O and 0.045% HCl) and buffer B (80%acetonitrile, 20% H2O and 0.045% HCl) was employed. HClwas used as the counter ion since TFA may affect ThioflavinT kinetic assays and cell toxicity experiments. The productfrom the HPLC purification was dissolved in 100%hexafluoroisopropanol (10 mg/ml) for 4 h and subjected to asecond round of HPLC purification to remove any residualscavengers. Direct injection electrospray MS confirmed themass of the pure peptide: expected 3903.32, observed3903.03. Analytical reverse-phase HPLC (C18) was used toconfirm the purity of the samples. Peptides were greater than96% pure. Quality control experiments confirmed that thepurified synthetic hIAPP formed amyloid fibres with typicalmorphology after aggregation in vitro as assessed by transmis-sion electron microscopy (electronic supplementary material[ESM] Fig. 1).

Thioflavin T assaysKinetics of hIAPP aggregation were deter-mined using solutions containing 20 μmol/l synthetic hIAPP,

32 μmol/l Thioflavin T (no. T3516; Sigma) and 20 mmol/lTris-HCl (pH 7.4) in the presence or absence of 4 pg/ml IL-1β(no. CYT-273; ProSpec) at 37°C with plate shaking every10 min and measurements taken every 60 min for 48 h.Thioflavin T is a small molecule, the quantum yield of whichincreases upon binding to amyloid fibres. The dye offers asimple probe of amyloid formation and has been widely usedin studies of amyloid formation by hIAPP [36, 37]. Theconcentration of Thioflavin T used has been shown not to alterthe kinetics of hIAPP amyloid formation [36]. Control reac-tions were carried out in the absence of hIAPP. Fluorescencemeasurements were recorded on a Beckman Coulter DTX880plate reader (Beckman Coulter, Brea, CA, USA) using anexcitation λ of 450 nm and emission λ of 485 nm. The assaywas performed in triplicate in four independent experiments.

Quantitative real-time RT-PCR Total RNAwas recovered fromislets using the High Pure RNA Isolation Kit (Roche, Basel,Switzerland), then reverse transcribed and subjected to quan-titative RT-PCR. Data were normalised to 18S rRNA levels

and expressed as fold relative to control using the 2−ΔΔCt

method. All quantitative RT-PCR data represent means oftriplicate determinations. Taqman probes (ThermoFisherScientific,Waltham,MA, USA)were used to quantify relativeexpression of the following mRNA targets: Atf4(Mm00515324); Cpe (Mm00516341_m1); Ddit3/Chop(Mm00492097_m1); Iapp (Mm00439403_m1); Il1b(Mm00434228_m1); Ins2 (Mm00731595_gH); Nos2(Mm00440502_m1); Pcsk1 (Mm00479023_m1); Pcsk2(Mm00500981_m1); and 18S rRNA (HS99999901_s1).

Insulin and IAPP assays After a culture period of 48 h in theindicated condition, islet media were collected and isletcontent extracted with acid–ethanol. IAPP was measured inmedia and cell extracts with an IAPP ELISA against total(amidated and deamidated) hIAPP with a capture antibodythat recognises an epitope near the midpoint of monomerichIAPP (no. EZHAT-51K; Millipore, Burlington, MA,USA). Insulin was measured with an insulin ELISA (no. 80-INSMSU-E10; ALPCO, Salem, NH, USA) [15]. ELISAexperimenters were blinded to the culture condition of isletsamples. Each measure represents the mean of triplicate deter-minations of five islets.

Statistical analyses

Amyloid prevalence was defined as the percentage of all isletsanalysed in each condition that contained amyloid. Amyloidseverity was defined as (Σ amyloid area/Σ islet area) × 100%for all islets from each condition, beta cell area was calculatedas (Σ beta cell area/Σ islet area) × 100% for all islets from eachcondition, and beta cell death was calculated as (Σ dead beta

2387Diabetologia (2020) 63:2385–2395

cells/Σ total cells) × 100% for all islets from each condition [1,17, 38]. Normality of data distribution in each dataset wastested using the D’Agostino–Pearson normality test. Pairedstatistics were applied when all replicates contained all exper-imental conditions and islets were isolated, cultured andanalysed at the same time within a replicate. For datasets withtwo groups, normally distributed data were analysed usingpaired two-tailed Student’s t test, while non-normally distrib-uted data were analysed using the Wilcoxon matched-pairssigned rank test. Datasets with more than two groups wereall found to be normally distributed, with ordinary one-wayANOVA used for unpaired data and repeated measures one-way ANOVA used for paired data. ANOVA tests werefollowed by Fisher’s least significant difference (LSD) posthoc tests to compare specific groups. Statistical analyses wereconducted with GraphPad Prism 8 software (GraphPad, SanDiego, CA, USA). Data are presented as mean ± SEM and avalue of p < 0.05 was considered significant.

Results

Low concentration IL-1β increases islet amyloid prevalenceand severity in vitro hIAPP-expressing islets were treated withrecombinant mouse IL-1β in concentrations ranging from 0 to4000 pg/ml to examine the dose–response effect on isletamyloid deposition. These concentrations of IL-1β werefound to increase islet amyloid deposition, with concentra-tions as low as 4 pg/ml significantly increasing amyloid prev-alence (Fig. 1a) and severity (Fig. 1b). A separate observerconfirmed these findings in an independent series of experi-ments using only the 4 pg/ml concentration of IL-1β, with thislow concentration again significantly increasing islet amyloidprevalence (93.52 ± 3.89% vs 43.83 ± 9.67% amyloid-containing islets, n = 7, p < 0.05) (Fig. 1c) and severity (4.45± 0.82% vs 2.16 ± 0.50% amyloid area/islet area, n = 7,p < 0.05) (Fig. 1d). Representative images of control(vehicle-treated) and IL-1β-treated hIAPP-expressing isletsillustrate the amyloid deposition observed (Fig. 1e).

Neutralisation of IL-1β prevents low concentration IL-1β-mediated islet amyloid deposition and beta cell loss To testthe specificity of the effect of low concentration (4 pg/ml) IL-1β treatment, studies using an IL-1β neutralising antibodywere undertaken. Low concentration IL-1β again increasedamyloid deposition (Fig. 2a, b), resulting in a significantreduction in beta cell area (58.47 ± 2.85% vs 64.37 ± 2.67%insulin area/islet area, n = 8, p < 0.05) (Fig. 2c) and a signifi-cant increase in beta cell death (0.567 ± 0.042% vs 0.346 ±0.021% dead beta cells/total cells, n = 8, p < 0.01) (Fig. 2d)compared with vehicle treatment. When IL-1β wasneutralised with an excess of IL-1β antibody (0.4 μg/ml),amyloid severity (Fig. 2b) and beta cell death (Fig. 2d) were

decreased, and beta cell area (Fig. 2c) was increased comparedwith IL-1β treatment alone. A non-significant (p = 0.08)reduction in amyloid prevalence was also observed when IL-1β was neutralised, compared with IL-1β alone (Fig. 2a). Onits own, the neutralising antibody showed no significant effecton these measures (Fig. 2a–d). Representative images illus-trate the amyloid deposition and beta cell area observed underthese conditions (Fig. 2e).

hIAPP aggregation kinetics are not increased by low concen-tration IL-1β in a cell-free system To determine whether theincreased amyloid deposition observed with low concentra-tion IL-1β was due to an effect of the cytokine to directlyinteract with hIAPP, cell-free Thioflavin T assays wereperformed to monitor hIAPP aggregation kinetics. The pres-ence of IL-1β did not change the rate of hIAPP aggregation ormaximal signal in the Thioflavin T fibril formation assays(Fig. 3a, b) over 48 h. Further, no aggregation was observedwith IL-1β or buffer in the absence of hIAPP (Fig. 3a, b).These findings suggest that IL-1β’s effect of increasing isletamyloid deposition is mediated by IL-1β signal transductionin islet cells.

Low concentration IL-1β increases IAPP release in vitro andthis effect is blocked by neutralisation of IL-1β To examinewhether increased IAPP release plays a role in IL-1β-mediatedamyloid deposition in our model, we quantified cumulativerelease of IAPP from islets cultured for 48 h in the presence orabsence of low concentration IL-1β. IL-1β at 4 pg/ml signifi-cantly increased IAPP (0.409 ± 0.082 vs 0.165 ± 0.051 pmol/5islets, n= 4, p < 0.05) (Fig. 4a) and insulin (87.5 ± 8.81 vs 48.3 ±17.3 pmol/5 islets, n= 4, p < 0.05) (Fig. 4b) release from isletsover 48 h in culture, but did not alter the IAPP/insulin secretionratio (data not shown). At the end of this culture period, neitherIAPP (1.14 ± 0.148 vs 1.17 ± 0.210 pmol/5 islets, n= 4) (Fig. 4c)nor insulin (84.0 ± 7.11 vs 72.1 ± 12.3 pmol/5 islets) (Fig. 4d)protein content was altered in IL-1β-treated islets. Iapp (Fig. 4e)and Ins2 (Fig. 4f) mRNA expression were significantly reducedby low concentration IL-1β following 48 h culture. Notably, IL-1β neutralisation prevented the increase in release of IAPP (Fig.5a) and insulin (Fig. 5b) observed with low concentration IL-1βtreatment.

Low concentration IL-1β does not alter the unfolded proteinresponse or processing enzyme gene expression in hIAPP-expressing islets Expression of genes in the pathwaysinvolved in the unfolded protein response (UPR) andproIAPP processing were examined to determine whetherthey may contribute to the observed effect of IL-1β onamyloid deposition in our model. Since increased Nos2 geneexpression is a marker of beta cell IL-1β signalling at phar-macological concentrations [39], we first confirmed that Nos2expression was increased in the presence of 4 pg/ml IL-1β.

2388 Diabetologia (2020) 63:2385–2395

We found Nos2 gene expression was increased more than 15-fold compared with when no IL-1β was present (Fig. 6a),indicating that IL-1β stimulates known cell signalling eventsat this concentration. To evaluate whether low concentrationIL-1β alters the UPR or proIAPP processing, we next exam-ined gene expression ofAtf4 andDdit3 as markers of the UPR,along with Pcsk1, Pcsk2 and Cpe as indicators of proIAPPprocessing. In contrast to Nos2, low concentration IL-1β hadno effect on islet expression of Atf4 (Fig. 6b) or Ddit3 (Fig.6c). Further, it did not alter gene expression of the proIAPPprocessing enzymes Pcsk1 (Fig. 6d), Pcsk2 (Fig. 6e) or Cpe(Fig. 6f).

Discussion

Study of the relationship between islet amyloid formation andIL-1β signalling has focused primarily on the role of hIAPP in

inducing IL-1β production from islet macrophages, thuspromoting islet inflammation and beta cell loss [18, 19].However, recent studies provide evidence that not only doeshIAPP aggregation lead to IL-1β production but, conversely,that islet IL-1β signalling promotes amyloid formation[21–23]. These findings suggest that a feed-forward systemof islet inflammation related to IL-1β signalling, hIAPPaggregation and beta cell cytotoxicity exists.

In this study we used an isolated islet model of endogenousbeta cell hIAPP expression and pg/ml concentrations of IL-1βto demonstrate, for the first time, that low concentration IL-1βstimulates IAPP release from islet beta cells, and that underconditions associated with amyloid formation in vitro, thisincreased release of IAPP leads to increased islet amyloiddeposition. Although previous studies demonstrated that lowconcentrations of IL-1β can stimulate insulin secretion[27–30, 40], its effect on IAPP release has not been reportedor linked to IL-1β-mediated islet amyloid formation and beta

Fig. 1 Low concentration IL-1βincreases islet amyloid prevalenceand severity in vitro. (a, b)Treatment of hIAPP-expressingmouse islets with IL-1β (0–4000 pg/ml) dose-dependentlyincreases islet amyloid prevalence(% of islets containing amyloid)(a) and amyloid severity (% ofamyloid-positive islet area) (b).(c, d) Induction of amyloidprevalence (c) and amyloidseverity (d) were verified inseparate experiments using onlyvehicle (black circles) and 4 pg/ml IL-1β (black squares). (e)Representative images of vehicle-and 4 pg/ml IL-1β-treatedhIAPP-expressing mouse islets(insulin, red; amyloid, green;DNA, blue). Scale bar, 50 μm.Data are presented as mean ±SEM. n = 3–7. *p < 0.05,**p < 0.01, ***p < 0.001. CTL,control (vehicle)

2389Diabetologia (2020) 63:2385–2395

Fig. 2 Neutralisation of IL-1βprevents low concentration IL-1β-mediated islet amyloiddeposition and beta cell loss. (a–d) hIAPP-expressing mouse isletstreated with low concentration IL-1β (4 pg/ml, black squares)exhibited increased amyloidprevalence (a) and severity (b),reduced beta cell area (c) andincreased beta cell death (d) whencompared with islets treated withvehicle (black circles). IL-1β-mediated changes in amyloidseverity, beta cell area and betacell death were blocked by co-treatment with an IL-1βneutralising antibody (0.4 μg/ml,white squares). (e) Representativeimages of hIAPP-expressingmouse islets treated in thepresence or absence of lowconcentration IL-1β and IL-1βneutralising antibody (insulin,red; amyloid, green; DNA, blue).Scale bar, 50 μm. Data arepresented as mean ± SEM. n = 8.*p < 0.05, **p < 0.01. CTL,control (vehicle); NAb, IL-1βneutralising antibody

Fig. 3 Low concentration IL-1β does not increase hIAPP aggregationkinetics. (a) Aggregation of hIAPPwas carried out in the absence (circles)or presence (squares) of 4 pg/ml IL-1β and monitored for 48 h byThioflavin T fluorescence in a cell-free system. One representative exper-iment is shown with technical replicates. (b) Quantification of maximal

hIAPP aggregation revealed no differences in hIAPP aggregation without(black circles) or with (black squares) 4 pg/ml IL-1β. In the absence ofhIAPP, neither buffer nor IL-1β increased Thioflavin T signal. Data arepresented as mean ± SEM. n = 4. AU, arbitrary units

2390 Diabetologia (2020) 63:2385–2395

cell loss. It is known that circulating IL-1β promotes insulinresistance, which in turn stimulates insulin secretion to main-tain euglycaemia in vivo [41, 42]. However, less is knownabout the mechanisms underlying the direct effect of lowconcentration IL-1β on IAPP and insulin release from isletsin vitro [27–29].

Given our in vitro approach using isolated islets and theexperimental conditions tested, we believe we have negatedany differences that may be imposed on the beta cell secretoryresponse by the varying glucose concentrations or differencesin insulin sensitivity that can occur in vivo. Therefore, ourfindings suggest that in vivo, physiologically relevant concen-trations of IL-1β derived from the circulation and/or islet

immune cells can directly induce IAPP release from the betacell and promote cytotoxic islet amyloid formation in thesetting of type 2 diabetes. Importantly, we believe that thisIL-1β-mediated increase in IAPP release is not likely to bethe factor responsible for initiating amyloid formation, butrather for exacerbating it. This hypothesis is supported bythe observations that IAPP is a normal beta cell secretoryproduct co-released with insulin [6] and that islet amyloidtypically is not observed in healthy humans [1] who exhibithigher IAPP release than those with type 2 diabetes [43].Additional studies are therefore needed to understand thefactors that initiate islet amyloid formation and allow IL-1βto exacerbate beta cell loss in the pathophysiology of type 2diabetes.

We identified increased release of IAPP by the beta cell,and thus increased substrate for amyloid formation, as a majormechanism underlying the effect of low concentration IL-1βon islet amyloid deposition. This finding is consistent withprevious studies that observed a positive relationship betweenbeta cell IAPP release and islet amyloid deposition [44, 45].Although concentrations of cytokines higher than those usedhere have been shown to induce the UPR [26, 46], we foundthat low concentration IL-1β treatment did not increaseexpression of the UPR markers Atf4 or Ddit3 in hIAPP-expressing islets, indicating that the observed increase inamyloid deposition was not associated with endoplasmicreticulum stress. Additionally, we found that low concentra-tion IL-1β did not change gene expression of the proIAPPprocessing enzymes Pcsk1, Psck2 or Cpe. Although IL-1βhas been shown to impair IAPP processing in the setting ofislet amyloid formation in other models [21], our data suggestthat low concentration IL-1β may not have this effect inmouse islets expressing hIAPP under the endogenous promot-er. Furthermore, our quantification of hIAPP aggregationkinetics in a cell-free system shows that IL-1β does not direct-ly increase hIAPP aggregation via physical interaction.Together, these data indicate that low concentration IL-1βsignalling leads to increased beta cell IAPP release and thatthis increase in substrate underlies IL-1β’s effect of increasingislet amyloid formation under pathophysiological conditions.

To evaluate the role of IL-1β signalling in promoting isletamyloid formation in situ, it is important to know the concen-tration of IL-1β that beta cells are exposed to under normaland pathophysiological conditions [25]. In the setting of meta-bolic disease, the level of IL-1β present in islets is thought todepend primarily on levels of circulating IL-1β originatingfrom immune cells in inflamed adipose tissue [24, 47, 48]and also on intra-islet production of IL-1β from islet immunecells such as macrophages [18–20]. Circulating concentra-tions of IL-1β in control mice are estimated to be 10 pg/ml,with an increase to ~35 pg/ml observed in obese db/db mice[24]. In humans, circulating IL-1β concentrations are thoughtto range from 0.5 to 12 pg/ml [24, 48], although some reports

Fig. 4 Low concentration IL-1β increases IAPP release in vitro. (a–d)Treatment of hIAPP-expressing mouse islets with 4 pg/ml IL-1β (blacksquares) increased release of IAPP (a) and insulin (b) into the media over48 h in culture when compared with vehicle treatment (black circles). Atthe end of the 48 h culture period, the islet content of IAPP (c) and insulin(d) protein did not differ between groups. (e, f) Iapp (e) and Ins2 (f)mRNA expression were decreased in IL-1β-treated (black squares)compared with vehicle-treated (black circles) islets (data normalised to18S rRNA levels and expressed as fold relative to control using the2−ΔΔCt method). Data are presented as mean ± SEM. n = 4–9.*p < 0.05, **p < 0.01

2391Diabetologia (2020) 63:2385–2395

suggest they are higher [47]. Estimating the concentration ofIL-1β inside the islet is difficult. However, it is believed thatcytokine production by islet immune cells, such as in hIAPP-induced IL-1β production bymacrophages, results in a furtherelevation of intra-islet cytokine concentrations and an increasein islet cytokine signalling [18, 19, 25]. Thus, we believe the

concentration of IL-1β used here approximates those found inthe islet microenvironment in vivo. As such, this study repli-cates the role of islet IL-1β signalling in promoting IAPPrelease and amyloid deposition in situ, and we believe it ismore informative for understanding islet pathophysiologythan studies using high concentrations of IL-1β.

Fig. 6 Low concentration IL-1β does not alter the UPR or processingenzyme gene expression in hIAPP-expressing islets. (a–c) Treatment ofhIAPP-expressing mouse islets with low concentration IL-1β for 48 h(black squares) significantly increased Nos2 mRNA expression,compared with vehicle treatment (black circles) (a), but mRNA expres-sion of Atf4 (b) orDdit3 (c), markers of the UPR, remained unaltered. (d–f) Similarly, mRNA expression of the proIAPP processing enzymes

Pcsk1 (d), Pcsk2 (e) and Cpe (f) was unchanged in hIAPP-expressingislets following vehicle treatment (black circles) vs treatment with 4 pg/ml IL-1β (black squares). Data are normalised to 18S rRNA levels andexpressed as fold relative to control using the 2−ΔΔCt method. Data arepresented as mean ± SEM. n = 5. **p < 0.01

Fig. 5 Neutralisation of IL-1β blocks low concentration IL-1β-mediatedIAPP release. Following 48 h treatment, release of IAPP (a) and insulin(b) from hIAPP-expressing islets were significantly increased by 4 pg/mlIL-1β (black squares) compared with vehicle (black circles). The releaseof IAPP (a) and insulin (b) mediated by low concentration IL-1β (black

squares) was reduced when islets were co-treated with 0.4μg/ml of IL-1βneutralising antibody (white squares). Data are presented as mean ± SEM.n = 4. *p < 0.05, **p < 0.01. CTL, control (vehicle); NAb, IL-1βneutralising antibody

2392 Diabetologia (2020) 63:2385–2395

Our findings suggest that islet IL-1β signalling andamyloid formation participate in a feed-forward system of isletinflammation and beta cell loss in the pathogenesis of type 2diabetes. We propose a model wherein beta cell IL-1β signal-ling increases hIAPP release and amyloid formation, as shownin the present study. In response to hIAPP aggregation, isletmacrophages increase NLRP3-dependent IL-1β production[18], further enhancing IL-1β signalling in the beta cell,hIAPP release, cytotoxic amyloid formation and beta cell loss[1, 3]. Over time, such a system of amyloid formation and isletinflammation may be an important contributor to the beta celldysfunction and loss observed in type 2 diabetes. It should benoted, in contrast to our model, that depletion of macrophagesin vivo has been shown to increase islet amyloid deposition[19], an observation consistent with earlier findings that isletamyloid fibrils can be phagocytosed by macrophages [49].Thus, macrophages likely play a dual role, removing amyloidfibrils to maintain islet homeostasis under physiologicalconditions, and increasing IL-1β production and hIAPPaggregation in pathophysiological settings related to type 2diabetes.

In summary, we have demonstrated that a concentration ofIL-1β approximating the concentration found in the isletin vivo increases islet amyloid formation and decreases betacell area. This effect is likely mediated through increasedrelease of hIAPP from the beta cell, thereby providing addi-tional substrate for cytotoxic amyloid formation.We posit thatthis effect of low concentration IL-1β to stimulate isletamyloid formation leads to a feed-forward system of IL-1βproduction, hIAPP aggregation and beta cell toxicity. Earlyefforts to block this effect of IL-1β may protect the islet frominflammation and hIAPP-induced beta cell cytotoxicity in thesetting of type 2 diabetes.

Acknowledgements We thank E. Boyko (University of Washington,Seattle, WA, USA) and S. Edelstein (George Washington University,Washington, DC, USA) for consultation related to statistical analyses.We thank B. Barrow, B. Fountaine, D. Hackney, S. Mongovin and C.Schmidt (Seattle Institute for Biomedical and Clinical Research, Seattle,WA, USA) for the excellent technical assistance provided during theperformance of these studies. We thank N. Eberhardt, (Mayo Clinic,Rochester, MN, USA) for providing mice with endogenous hIAPPexpression for use in this study. Some of the data from this study werepresented at the ADA77th and 79th Scientific Sessions in 2017 and 2019.The graphical abstract was created with BioRender.com.

Data availability Data supporting the conclusions of this work are includ-ed within the article and are available from the corresponding author onreasonable request.

Funding This work was supported by funding from the US Departmentof Veterans Affairs (IK2BX004659 to ATT, I01BX004063 to RLH,I01BX001060 to SEK), National Institutes of Health (P30DK017047 tothe University of Washington Diabetes Research Center, T32DK007247and T32HL007028 to the University of Washington, F32DK109584 toMFH, R01DK098506 to SZ, R01GM078114 to DPR), AmericanDiabetes Association (Mentor-Based Fellowship to SEK, Postdoctoral

Fellowship 1-18-PDF-174 to MFH), the University of Washington(McAbee Fellowship to DTM, MFH and NE), the French Society ofDiabetes (to NE) and the Swiss National Science Foundation (to DTM).

Duality of interest The authors declare that there are no relationships oractivities that might bias, or be perceived to bias, this work.

Contribution statement ATT participated in study design, performedresearch, analysed and interpreted data and wrote/approved the manu-script. MM, DTM, NE and MFH participated in study design, performedresearch, analysed and interpreted data, and revised/approved the manu-script. RA and DPR supplied critical reagents, performed research,analysed and interpreted data, and revised/approved the manuscript.JJC, SZ, RLH and SEK participated in study design, analysed andinterpreted data, and revised/approved the manuscript. ATT and SEKare responsible for the integrity of the work as a whole.

References

1. Jurgens CA, Toukatly MN, Fligner CL et al (2011) β-Cell loss andβ-cell apoptosis in human type 2 diabetes are related to isletamyloid deposition. Am J Pathol 178(6):2632–2640. https://doi.org/10.1016/j.ajpath.2011.02.036

2. Zhao H-L, Lai FMM, Tong PCY et al (2003) Prevalence and clin-icopathological characteristics of islet amyloid in Chinese patientswith type 2 diabetes. Diabetes 52(11):2759–2766. https://doi.org/10.2337/diabetes.52.11.2759

3. Westermark P (1972) Quantitative studies on amyloid in the isletsof Langerhans. Ups J Med Sci 77(2):91–94. https://doi.org/10.1517/03009734000000014

4. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC(2003) β-Cell deficit and increased beta-cell apoptosis in humanswith type 2 diabetes. Diabetes 52(1):102–110. https://doi.org/10.2337/diabetes.52.1.102

5. MacArthur DL, de Koning EJ, Verbeek JS, Morris JF, Clark A(1999) Amyloid fibril formation is progressive and correlates withbeta-cell secretion in transgenic mouse isolated islets. Diabetologia42(10):1219–1227. https://doi.org/10.1007/s001250051295

6. Kahn SE, D’Alessio DA, Schwartz MW et al (1990) Evidence ofcosecretion of islet amyloid polypeptide and insulin by β-cells.Diabetes 39(5):634–638. https://doi.org/10.2337/diab.39.5.634

7. Westermark P, Wernstedt C, Wilander E, Hayden DW, O’BrienTD, Johnson KH (1987) Amyloid fibrils in human insulinomaand islets of Langerhans of the diabetic cat are derived from aneuropeptide-like protein also present in normal islet cells. ProcNatl Acad Sci U S A 84(11):3881–3885. https://doi.org/10.1073/pnas.84.11.3881

8. Westermark P, Wilander E, Westermark GT, Johnson KH (1987)Islet amyloid polypeptide-like immunoreactivity in the islet B cellsof type 2 (non-insulin-dependent) diabetic and non-diabetic indi-viduals. Diabetologia 30(11):887–892. https://doi.org/10.1007/BF00274799

9. Cooper GJ, Willis AC, Clark A, Turner RC, Sim RB, Reid KB(1987) Purification and characterization of a peptide fromamyloid-rich pancreases of type 2 diabetic patients. Proc NatlAcad Sci U S A 84(23):8628–8632. https://doi.org/10.1073/pnas.84.23.8628

10. Westermark P, Engström U, Johnson KH, Westermark GT,Betsholtz C (1990) Islet amyloid polypeptide: pinpointing aminoacid residues linked to amyloid fibril formation. Proc Natl Acad SciU S A 87(13):5036–5040

2393Diabetologia (2020) 63:2385–2395

11. D’Alessio DA, Verchere CB, Kahn SE et al (1994) Pancreaticexpression and secretion of human islet amyloid polypeptide in atransgenic mouse. Diabetes 43(12):1457–1461. https://doi.org/10.2337/diab.43.12.1457

12. Butler AE, Jang J, Gurlo T, Carty MD, Soeller WC, Butler PC(2004) Diabetes due to a progressive defect in β-cell mass in ratstransgenic for human islet amyloid polypeptide (HIP rat): a newmodel for type 2 diabetes. Diabetes 53(6):1509–1516. https://doi.org/10.2337/diabetes.53.6.1509

13. Soeller WC, Janson J, Hart SE et al (1998) Islet amyloid-associateddiabetes in obese A(vy)/a mice expressing human islet amyloidpolypeptide. Diabetes 47(5):743–750. https://doi.org/10.2337/diabetes.47.5.743

14. Hiddinga HJ, Sakagashira S, IshigameM et al (2012) Expression ofwild-type and mutant S20G hIAPP in physiologic knock-in mousemodels fails to induce islet amyloid formation, but induces mildglucose intolerance. J Diabetes Investig 3(2):138–147. https://doi.org/10.1111/j.2040-1124.2011.00166.x

15. Meier DT, Entrup L, Templin AT et al (2016) The S20G substitu-tion in hIAPP is more amyloidogenic and cytotoxic than wild-typehIAPP in mouse islets. Diabetologia 59(10):2166–2171. https://doi.org/10.1007/s00125-016-4045-x

16. Kahn SE, Andrikopoulos S, Verchere CB (1999) Islet amyloid: along-recognized but underappreciated pathological feature of type 2diabetes. Diabetes 48(2):241–253. https://doi.org/10.2337/diabetes.48.2.241

17. Hull RL, Andrikopoulos S, Verchere CB et al (2003) Increaseddietary fat promotes islet amyloid formation and β-cell secretorydysfunction in a transgenic mouse model of islet amyloid. Diabetes52(2):372–379. https://doi.org/10.2337/diabetes.52.2.372

18. Masters SL, Dunne A, Subramanian SL et al (2010) Activation ofthe NLRP3 inflammasome by islet amyloid polypeptide provides amechanism for enhanced IL-1β in type 2 diabetes. Nat Immunol11(10):897–904. https://doi.org/10.1038/ni.1935

19. Westwell-Roper CY, Ehses JA, Verchere CB (2014) Residentmacrophages mediate islet amyloid polypeptide-induced islet IL-1β production and β-cell dysfunction. Diabetes 63(5):1698–1711.https://doi.org/10.2337/db13-0863

20. Meier DT, Morcos M, Samarasekera T, Zraika S, Hull RL, KahnSE (2014) Islet amyloid formation is an important determinant forinducing islet inflammation in high-fat-fed human IAPP transgenicmice. Diabetologia 57(9):1884–1888. https://doi.org/10.1007/s00125-014-3304-y

21. Park YJ, Warnock GL, Ao Z et al (2017) Dual role of interleukin-1β in islet amyloid formation and its β-cell toxicity: implicationsfor type 2 diabetes and islet transplantation. Diabetes Obes Metab19(5):682–694. https://doi.org/10.1111/dom.12873

22. Hui Q, Asadi A, Park YJ et al (2017) Amyloid formation disruptsthe balance between interleukin-1β and interleukin-1 receptorantagonist in human islets. Mol Metab 6(8):833–844. https://doi.org/10.1016/j.molmet.2017.05.016

23. Westwell-Roper CY, Chehroudi CA, Denroche HC, Courtade JA,Ehses JA, Verchere CB (2015) IL-1 mediates amyloid-associatedislet dysfunction and inflammation in human islet amyloid polypep-tide transgenic mice. Diabetologia 58(3):575–585. https://doi.org/10.1007/s00125-014-3447-x

24. O’Neill CM, Lu C, Corbin KL et al (2013) Circulating levels of IL-1B+IL-6 cause ER stress and dysfunction in islets from prediabeticmale mice. Endocrinology 154(9):3077–3088. https://doi.org/10.1210/en.2012-2138

25. Nunemaker CS (2016) Considerations for defining cytokine dose,duration, and milieu that are appropriate for modeling chronic low-grade inflammation in type 2 diabetes. J Diabetes Res 2016:2846570. https://doi.org/10.1155/2016/2846570

26. Templin AT, Maier B, Tersey SA, Hatanaka M, Mirmira RG(2014) Maintenance of Pdx1 mRNA translation in islet β-cells

during the unfolded protein response. Mol Endocrinol 28(11):1820–1830. https://doi.org/10.1210/me.2014-1157

27. Palmer JP, Helqvist S, Spinas GA et al (1989) Interaction of β-cellactivity and IL-1 concentration and exposure time in isolated ratislets of Langerhans. Diabetes 38(10):1211–1216. https://doi.org/10.2337/diab.38.10.1211

28. Jeong I-K, Oh S-H, Chung J-H et al (2002) The stimulatory effectof IL-1β on the insulin secretion of rat pancreatic islet is not relatedwith iNOS pathway. ExpMolMed 34(1):12–17. https://doi.org/10.1038/emm.2002.2

29. Arous C, Ferreira PG, Dermitzakis ET, Halban PA (2015) Shortterm exposure of beta cells to low concentrations of interleukin-1βimproves insulin secretion through focal adhesion and actin remod-eling and regulation of gene expression. J Biol Chem 290(10):6653–6669. https://doi.org/10.1074/jbc.M114.611111

30. Hajmrle C, Smith N, Spigelman AF et al (2016) Interleukin-1signaling contributes to acute islet compensation. JCI Insight 1(4):e86055. https://doi.org/10.1172/jci.insight.86055

31. Zraika S, Hull RL, Udayasankar J et al (2007) Glucose- and time-dependence of islet amyloid formation in vitro. Biochem BiophysRes Commun 354(1):234–239. https://doi.org/10.1016/j.bbrc.2006.12.187

32. Bower RL, Yule L, Rees TA et al (2018) Molecular signature forreceptor engagement in the metabolic peptide hormone amylin.ACS Pharmacol Transl Sci 1(1):32–49. https://doi.org/10.1021/acsptsci.8b00002

33. Tu L-H, Serrano AL, Zanni MT, Raleigh DP (2014) Mutationalanalysis of preamyloid intermediates: the role of his-tyr interactionsin islet amyloid formation. Biophys J 106(7):1520–1527. https://doi.org/10.1016/j.bpj.2013.12.052

34. Marek P, Woys AM, Sutton K, Zanni MT, Raleigh DP (2010)Efficient microwave-assisted synthesis of human islet amyloidpolypeptide designed to facilitate the specific incorporation oflabeled amino acids. Org Lett 12(21):4848–4851. https://doi.org/10.1021/ol101981b

35. Abedini A, Raleigh DP (2005) Incorporation of pseudoprolinederivatives allows the facile synthesis of human IAPP, a highlyamyloidogenic and aggregation-prone polypeptide. Org Lett 7(4):693–696. https://doi.org/10.1021/ol047480+

36. Tu L-H, Raleigh DP (2013) Role of aromatic interactions inamyloid formation by islet amyloid polypeptide. Biochemistry52(2):333–242. https://doi.org/10.1021/bi3014278

37. LeVine H (1999) Quantification of beta-sheet amyloid fibril struc-tures with thioflavin T. Methods Enzymol 309:274–284

38. Wang F, Hull RL, Vidal J, Cnop M, Kahn SE (2001) Islet amyloiddevelops diffusely throughout the pancreas before becoming severeand replacing endocrine cells. Diabetes 50(11):2514–2520. https://doi.org/10.2337/diabetes.50.11.2514

39. Eizirik DL, FlodströmM, Karlsen AE,Welsh N (1996) The harmo-ny of the spheres: inducible nitric oxide synthase and related genesin pancreatic beta cells. Diabetologia 39(8):875–890. https://doi.org/10.1007/BF00403906

40. Donath MY, Böni-Schnetzler M, Ellingsgaard H, Halban PA,Ehses JA (2010) Cytokine production by islets in health and diabe-tes: cellular origin, regulation and function. Trends EndocrinolMetab 21(5):261–267. https://doi.org/10.1016/j.tem.2009.12.010

41. Stienstra R, Joosten LAB, Koenen T et al (2010) Theinflammasome-mediated caspase-1 activation controls adipocytedifferentiation and insulin sensitivity. Cell Metab 12(6):593–605.https://doi.org/10.1016/j.cmet.2010.11.011

42. Vandanmagsar B, Youm Y-H, Ravussin A et al (2011 Feb) TheNLRP3 inflammasome instigates obesity-induced inflammationand insulin resistance. Nat Med 17(2):179–188. https://doi.org/10.1038/nm.2279

43. Kahn SE, Verchere CB, Andrikopoulos S et al (1998) Reducedamylin release is a characteristic of impaired glucose tolerance

2394 Diabetologia (2020) 63:2385–2395

and type 2 diabetes in Japanese Americans. Diabetes 47(4):640–645. https://doi.org/10.2337/diabetes.47.4.640

44. Hull RL, Shen Z-P, Watts MR et al (2005) Long-term treatmentwith rosiglitazone andmetformin reduces the extent of, but does notprevent, islet amyloid deposition in mice expressing the gene forhuman islet amyloid polypeptide. Diabetes 54(7):2235–2244.https://doi.org/10.2337/diabetes.54.7.2235

45. Aston-Mourney K, Subramanian SL, Zraika S et al (2013) One yearof sitagliptin treatment protects against islet amyloid-associated β-cell loss and does not induce pancreatitis or pancreatic neoplasia inmice. Am J Physiol Endocrinol Metab 305(4):E475–E484. https://doi.org/10.1152/ajpendo.00025.2013

46. CardozoAK, Ortis F, Storling J, FengY-M, Rasschaert J, TonnesenM et al (2005) Cytokines downregulate the sarcoendoplasmic retic-ulum pump Ca2+ ATPase 2b and deplete endoplasmic reticulumCa2+, leading to induction of endoplasmic reticulum stress inpancreatic β-cells. Diabetes 54(2):452–461. https://doi.org/10.2337/diabetes.54.2.452

47. Cannon JG, van der Meer JW, Kwiatkowski D et al (1988)Interleukin-1β in human plasma: optimization of blood collection,plasma extraction, and radioimmunoassay methods. LymphokineRes 7(4):457–467

48. Testa M, Yeh M, Lee P et al (1996) Circulating levels of cytokinesand their endogenous modulators in patients with mild to severecongestive heart failure due to coronary artery disease or hyperten-sion. J Am Coll Cardiol 28(4):964–971. https://doi.org/10.1016/s0735-1097(96)00268-9

49. Badman MK, Pryce RA, Chargé SBP, Morris JF, Clark A (1998)Fibrillar islet amyloid polypeptide (amylin) is internalised bymacrophages but resists proteolytic degradation. Cell Tissue Res291(2):285–294. https://doi.org/10.1007/s004410050998

Publisher’s note Springer Nature remains neutral with regard to jurisdic-tional claims in published maps and institutional affiliations.

2395Diabetologia (2020) 63:2385–2395